Model-based evolutionary algorithm for optimization of gas distribution systems in power plant electrostatic precipitators

摘要

Electrostatic precipitators (ESP) are used for dust separation at various industrial processes, e.g. from flue gas streams of coal-fired power plants. The efficiency of an ESP depends mainly on the gas velocity distribution within the separation zones. To ensure a sufficient flow field, a gas distribution system (GDS) is installed in the ESP inlet hood. Such a GDS consists of several hundred elements, e.g., perforated plates with different free cross sections, deflecting baffles and blocking plates, which leads to a very high number of possible GDS configurations. Usually a heuristic approach in conjunction with CFD simulations and cold flow measurements by hand is used to find a suitable GDS configuration. This results in an acceptable but not optimal gas velocity distribution especially at the inlet area of the first separation field. To find a better or even an optimal GDS configuration, a detailed CFD model of an ESP and a computational intelligent and automated approach based on an evolutionary algorithm (EA) is developed. The development process and first results are presented in this article.